Cleanup radiation tendencies, add RRTMGP tests on Cheyenne

[climbfuji]

This PR addresses issue NCAR/ccpp-physics#179. It only deals with the clean up of the standard names as described in the issue, it does not remove the LTP (extra layers for radiation) logic.

Regression tests for RRTMGP (in PROD, REPRO and DEBUG mode) and other tests in DEBUG mode are added for Cheyenne. Note that for the GNU compiler, the GSD DEBUG tests are currently disabled (all crashing).

For GNU, the new tests are simply added to the end of rt_gnu.conf, since it is expected that this file will be rewritten once the GNU configuration on hera works.

Associated PRs:

NCAR/ccpp-physics#422
NCAR/fv3atm#34
#32

For regression testing information, see below.

[climbfuji]

Regression tests against existing baseline on hera.intel all pass.

rt_hera_intel.log

[climbfuji]

Regression testing on cheyenne.gnu:

(1) Verify against existing baseline: all existing tests pass, all new tests run to completion but "fail" because of missing baselines (this is the expected outcome).

rt_cheyenne_gnu_against_existing_baseline.log

(2) Baselines are generated for the newly added tests and merged into the existing baseline (date tag 20200319) - this is no problem, because the existing baseline results are not modified.

rt_cheyenne_gnu_create_new_baseline.log

(3) Verification against the new/extended baseline: all tests pass.

rt_cheyenne_gnu_verify_against_new_baseline.log

[climbfuji]

Regression testing on cheyenne.intel:

(1) Verify against existing baseline: all existing tests pass, all new tests run to completion but "fail" because of missing baselines (this is the expected outcome).

rt_cheyenne_intel_against_existing_baseline.log

(2) Baselines are generated for the newly added tests and merged into the existing baseline (date tag 20200319) - this is no problem, because the existing baseline results are not modified.

rt_cheyenne_intel_create_new_baseline.log

(3) Verification against the new/extended baseline: all tests pass.

rt_cheyenne_intel_verify_against_new_baseline.log

[climbfuji]

@grantfirl @llpcarson I'd appreciate your eyes on this set of PRs. Should be quick and straightforward.

@grantfirl are there any other changes you would like to make, in particular address some of the RRTMGP issues you noted when we merged the code first, before we send all this to the authoritative repositories/branches?

[llpcarson]

approved

[grantfirl]

approved.